Re: [PATCH v11 7/9] coresight: etm4x: exclude ss_status from drvdata->config

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] coresight: data race and state leak in shared ss_status array
- [Medium] coresight: contradictory bitmask change introduces unhandled behavior
--

commit 1a408a1bbc8676e137115f3c8798cfbd8a499e15
Author: Yeoreum Yun <yeoreum.yun@xxxxxxx>

coresight: etm4x: exclude ss_status from drvdata->config
    
This patch moves the ss_status array out of the session-specific configuration 
structures and into the shared etmv4_drvdata structure. It aims to preserve 
existing behavior regarding single-shot comparator events while cleaning up 
the configuration data.

> diff --git a/drivers/hwtracing/coresight/coresight-etm4x-core.c b/drivers/hwtracing/coresight/coresight-etm4x-core.c
> index 70a5c0b4fad1c..9de0c268ff954 100644
> --- a/drivers/hwtracing/coresight/coresight-etm4x-core.c
> +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c
[ ... ]
> @@ -573,11 +573,11 @@ static int etm4_enable_hw(struct etmv4_drvdata *drvdata)
>  		etm4x_relaxed_write32(csa, config->res_ctrl[i], TRCRSCTLRn(i));
>  
>  	for (i = 0; i < caps->nr_ss_cmp; i++) {
> -		/* always clear status bit on restart if using single-shot */
> +		/* always clear status and pending bits on restart if using single-shot */
>  		if (config->ss_ctrl[i] || config->ss_pe_cmp[i])
> -			config->ss_status[i] &= ~TRCSSCSRn_STATUS;
> +			drvdata->ss_status[i] &= ~(TRCSSCSRn_STATUS | TRCSSCSRn_PENDING);

[Severity: Medium]
The commit message states that this patch intentionally preserves the existing
behavior regarding the clearing of PENDING and STATUS bits. However, the
pre-patch code only cleared TRCSSCSRn_STATUS. By explicitly adding
TRCSSCSRn_PENDING to the bitmask here, does this code actively introduce the
behavioral change it claims to defer?

[ ... ]
> @@ -1081,7 +1081,7 @@ static void etm4_disable_hw(struct etmv4_drvdata *drvdata)
>  
>  	/* read the status of the single shot comparators */
>  	for (i = 0; i < caps->nr_ss_cmp; i++) {
> -		config->ss_status[i] =
> +		drvdata->ss_status[i] =
>  			etm4x_relaxed_read32(csa, TRCSSCSRn(i));
>  	}

[Severity: High]
When a perf session stops, etm4_disable_hw() writes the hardware state
directly into the shared drvdata->ss_status array without holding
drvdata->spinlock.

Since a concurrent sysfs session could be writing to this same array in
sshot_ctrl_store() under the spinlock, can this lockless write cause a data
race?

Additionally, does this overwrite leak the perf session's single-shot state
into the shared array, allowing a subsequent sysfs session to inadvertently
observe or re-inject the leftover perf state?

> diff --git a/drivers/hwtracing/coresight/coresight-etm4x-sysfs.c b/drivers/hwtracing/coresight/coresight-etm4x-sysfs.c
> index f6d053f73264c..50ef2e957300f 100644
> --- a/drivers/hwtracing/coresight/coresight-etm4x-sysfs.c
> +++ b/drivers/hwtracing/coresight/coresight-etm4x-sysfs.c
[ ... ]
> @@ -1833,8 +1833,8 @@ static ssize_t sshot_ctrl_store(struct device *dev,
>  	raw_spin_lock(&drvdata->spinlock);
>  	idx = config->ss_idx;
>  	config->ss_ctrl[idx] = FIELD_PREP(TRCSSCCRn_SAC_ARC_RST_MASK, val);
> -	/* must clear bit 31 in related status register on programming */
> -	config->ss_status[idx] &= ~TRCSSCSRn_STATUS;
> +	/* must clear bit 31 and 30 in related status register on programming */
> +	drvdata->ss_status[idx] &= ~(TRCSSCSRn_STATUS | TRCSSCSRn_PENDING);
>  	raw_spin_unlock(&drvdata->spinlock);

[Severity: High]
As seen here, sysfs modifies the same shared array under the spinlock.
Is it safe to move ss_status to the shared drvdata struct without
synchronizing the hardware stop paths or segregating the session state?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260915-separate_etm_cfg_v2-v11-0-d2b258d51747@xxxxxxx?part=7